Karger et al. BMC Microbiology 2012, 12:229 http://www.biomedcentral.com/1471-2180/12/229

RESEARCH ARTICLE Open Access Rapid identification of Burkholderia mallei and Burkholderia pseudomallei by intact cell Matrix-assisted Laser Desorption/Ionisation mass spectrometric typing Axel Karger1, Rüdiger Stock2, Mario Ziller3, Mandy C Elschner4, Barbara Bettin1, Falk Melzer4, Thomas Maier5, Markus Kostrzewa5, Holger C Scholz2, Heinrich Neubauer4 and Herbert Tomaso4*

Abstract Background: Burkholderia (B.) pseudomallei and B. mallei are genetically closely related species. B. pseudomallei causes melioidosis in humans and animals, whereas B. mallei is the causative agent of glanders in equines and rarely also in humans. Both agents have been classified by the CDC as priority category B biological agents. Rapid identification is crucial, because both agents are intrinsically resistant to many antibiotics. Matrix-assisted laser desorption/ionisation mass spectrometry (MALDI-TOF MS) has the potential of rapid and reliable identification of pathogens, but is limited by the availability of a database containing validated reference spectra. The aim of this study was to evaluate the use of MALDI-TOF MS for the rapid and reliable identification and differentiation of B. pseudomallei and B. mallei and to build up a reliable reference database for both organisms. Results: A collection of ten B. pseudomallei and seventeen B. mallei strains was used to generate a library of reference spectra. Samples of both species could be identified by MALDI-TOF MS, if a dedicated subset of the reference spectra library was used. In comparison with samples representing B. mallei, higher genetic diversity among B. pseudomallei was reflected in the higher average Eucledian distances between the mass spectra and a broader range of identification score values obtained with commercial software for the identification of microorganisms. The type strain of B. pseudomallei (ATCC 23343) was isolated decades ago and is outstanding in the spectrum-based dendrograms probably due to massive methylations as indicated by two intensive series of mass increments of 14 Da specifically and reproducibly found in the spectra of this strain. Conclusions: Handling of pathogens under BSL 3 conditions is dangerous and cumbersome but can be minimized by inactivation of bacteria with ethanol, subsequent protein extraction under BSL 1 conditions and MALDI-TOF MS analysis being faster than nucleic amplification methods. Our spectra demonstrated a higher homogeneity in B. mallei than in B. pseudomallei isolates. As expected for closely related species, the identification process with MALDI Biotyper software (Bruker Daltonik GmbH, Bremen, Germany) requires the careful selection of spectra from reference strains. When a dedicated reference set is used and spectra of high quality are acquired, it is possible to distinguish both species unambiguously. The need for a careful curation of reference spectra databases is stressed.

* Correspondence: [email protected] 4Friedrich-Loeffler-Institut, Federal Research Institute for Animal Health, Institute of Bacterial Infections and Zoonoses, Naumburger Str. 96 a, Jena 07743, Germany Full list of author information is available at the end of the article

© 2012 Karger et al.; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Karger et al. BMC Microbiology 2012, 12:229 Page 2 of 15 http://www.biomedcentral.com/1471-2180/12/229

Background cultures, but no commercial tests are available [8,9]. Burkholderia (B.) pseudomallei and B. mallei are genet- Real-time PCR systems have been developed for diag- ically closely related bacterial species that can cause fatal nosing and differentiating as rapid alternatives to bio- disease in humans and animals. B. pseudomallei is a fac- chemical tests, but few have been validated on clinical ultative intracellular soil bacterium and the cause of samples [10-13]. melioidosis, which has the highest prevalence in the hot Matrix-assisted laser desorption ionization-time of and humid regions of Southeast Asia, and Northern flight mass spectrometry (MALDI-TOF MS) for identifi- Australia. The infection can be acquired by contact with cation of bacteria has become a useful tool for the rapid contaminated soil or water by inhalation or percutan- identification of bacteria (see [14] for a recent review). eously. Human infections can be asymptomatic, can In some studies intact cell mass spectrometry (ICMS) present with focal lesions that may affect almost any showed better correlation to genetic markers than con- organ, or may be disseminate resulting in pneumonia ventional morphological classification [15]. The tech- and septicaemia. In certain regions of Asia melioidosis is nique and modified procedures including a digestion a major cause of human morbidity and acute systemic step (‘shotgun mass mapping’) have been successfully melioidosis has a case fatality rate of up to 50% even if used for subspecies-level classification in some species treated [1,2]. Melioidosis has been described in wild ani- [16]. Characteristic features of ICMS are simple sample mals, but also in farm and pet animals and can be spread preparation procedures of whole cells, spectrum acquisi- by animal trade and transport [3]. Both species are tion in the mass range between approximately 2,000 and pathovars of a single genomospecies which was divided 15,000 Da and analysis based upon comparison of sam- historically in two separate species due to their clinical ple spectra with reference spectra. By statistical impact and host tropism. B. thailandensis is the third approaches, similarity between mass spectra can be closely related species of the so-called “Pseudomallei exploited for the identification of microorganisms. complex” which has been out-grouped from the species MALDI-TOF MS was also established for identification B. pseudomallei based on arabinose fermentation and its of non-fermenting gram-negative bacteria isolated from markedly lower pathogenicity. B. thailandensis and B. cystic fibrosis patients in Brazil [17]. Patients with cystic pseudomallei are soil bacteria that share the same geo- fibrosis suffer primarily under infections with Pseudo- graphical distribution. monads, but Burkholderiae play also an important role. B. mallei is a gram-negative, non-motile obligate In the Brazilian study a comprehensive number of Bur- pathogen and the causative agent of glanders and farcy kholderia species was included and could be identified in equines (horses, donkeys, mules). In horses, glanders correctly in most cases. However, neither B. pseudomal- primarily presents with purulent nasal discharge, inflam- lei nor B. mallei were among the clinical isolates tested. mation of the mucous membranes of the upper respira- Sporadic cases of melioidosis in cystic fibrosis patients tory tract, and poor general condition, whereas farcy is a have been described in the literature and seem to be an chronic cutaneous disease with formation of nodules emerging problem [18-22]. Due to increased travel activ- which may develop into ulcers. Equines are the only ity, international trade, climate change, and the potential known reservoir. Contact with infected animals, inges- threat of bioterrorist attacks infections caused by B. tion of glanderous meat and exposure to aerosols can pseudomallei and B. mallei can become a serious cause B. mallei infections in humans. Human glanders is problem. highly lethal and resembles melioidosis. Chronic and la- The aim of this study was to evaluate the potential tent infections can exacerbate into the acute form even benefit of MALDI-TOF MS for the rapid and reliable after 15 years in both diseases. Both bacterial species are identification and differentiation of B. pseudomallei and intrinsically resistant to many antibiotics including ampi- B. mallei. cillin and broad- and expanded-spectrum cephalospor- ines due to the production of a beta-lactamase [4]. B. Results mallei and B. pseudomallei have been classified by the Construction of a reference database CDC as priority category B biological agents. A custom made set of 34 reference spectra, which are Isolation and microbiological identification of B. pseu- called main spectra (MSP) in the MALDI Biotyper ter- domallei and B. mallei from clinical samples can take up minology (Bruker Daltonik GmbH, Bremen, Germany), to one week. Commercial biochemical test systems for was generated and used as the basis for all further calcu- B. mallei are not available and B. pseudomallei may be lations. This reference spectra set included all strains misidentified as Chromobacterium violaceum or other listed in Table 1 (B. mallei and B. pseudomallei) and bacteria [5-7]. Latex agglutination using a monoclonal additionally samples from B. ambifaria (DSM 16087), B. antibody was shown to be a valuable technique for the cenocepacia (ATCC BAA-245), B. dolosa (DSM 16088), rapid identification of B. pseudomallei in positive blood B. glathei (ATCC 29195), B. multivorans (DSM 13243), Karger et al. BMC Microbiology 2012, 12:229 Page 3 of 15 http://www.biomedcentral.com/1471-2180/12/229

Table 1 Burkholderia (B.) mallei and B. pseudomallei strains Bacteria Origin Country Year fliC fliP Motility B. mallei ATCC 23344T Human China 1942 + + - NCTC 120 unknown United Kingdom 1920 + + - NCTC 10230 Horse Hungary 1961 + + - NCTC 10247* Human Turkey 1960 + + - NCTC 10260 Human Turkey 1949 + + - M1* unknown unknown unknown + + - M2 unknown unknown unknown + + - Rotz7 (SVP) unknown unknown unknown + + - 32 unknown unknown unknown + + - 34 unknown unknown unknown + + - 235 unknown unknown unknown + + - 237 unknown unknown unknown + + - 242 unknown unknown unknown + + - Bogor Horse Indonesia unknown + + - Mukteswar Horse India unknown + + - Zagreb Horse former-Yugoslavia unknown + + - Dubai 7 Horse United Arab Emirates 2004 + + - B. pseudomallei ATCC 23343T Human unknown <1957 + - + EF 15660* unknown unknown unknown + - + NCTC 1688* Rat Malaysia 1923 + - + PITT 225A* Human Thailand 1986 + - + PITT 521 Human Pakistan 1988 + - + PITT 5691 unknown unknown unknown + - + 120107RR0019 Human Italy 2007 + - + H05410-0490 Human Asia unknown + - + 03-04448 Human unknown unknown + - + 03-04450 unknown unknown unknown + - + T type strain. *Constituents of the reduced reference set dedicated for the discrimination of B. mallei and B. pseudomallei. Characteristics of Burkholderia (B.) mallei and B. pseudomallei strains used to establish the database for the identification and differentiation with MALDI-TOF mass spectrometry. Species identity was confirmed by real-time PCR assays targeting a sequence of the fliC gene that is specific for both species but does not discriminate B. mallei from B. pseudomallei. The real-time PCR assay targeting fliP is specific for B. mallei. Motility was also assessed as a phenotypic marker because B. pseudomallei is motile while B. mallei is not.

B. stabilis (DSM 16586), and B. thailandensis (ATCC present in the MSP of the spectra collection by agglom- 700388). This set of 34 samples will be referred to as the eration of scores for every mass and performing a nor- ‘custom reference set’. The full set of MALDI Biotyper malisation that will result in a final value between 0 reference database entries will be referred to as ‘MALDI (unrelated) and 1000 (identical). For convenience, the re- Biotyper reference set’. In a first analysis, spectra of the sult is given as logarithmic value smaller than or equal custom reference set were queried against a combined to 3.0. Masses of the tested spectrum will be scored in a database composed of the custom reference set of 34 weighted fashion depending on their location within a Burkholderia samples and the MALDI Biotyper refer- narrower or a wider mass tolerance window centred on ence set. For every queried spectrum, MALDI Biotyper the masses of the MSP. Additionally, the score for every software generates a score-based ranked list of organ- coinciding mass of the tested spectrum will be weighted isms. The organism with the highest score is ranked first according to the frequency with which the correspond- (‘top hit’) and its species is taken as the result of the ing mass of the MSP has been found in the single spec- query. MSP scores are calculated by an algorithm that tra that were used for the construction of the MSP. compares the masses of a tested sample with those Thus, scores carry information on the number of Karger et al. BMC Microbiology 2012, 12:229 Page 4 of 15 http://www.biomedcentral.com/1471-2180/12/229

coinciding masses found in the tested spectrum and the Biotyper was repeated with single spectra of the B. mal- MSP, the mass aberration that is observed between the lei and B. pseudomallei samples from Table 1. The corresponding masses of the tested spectrum and the results were very similar to those obtained with MSP. MSP and the reproducibility of the respective masses of For B. mallei samples, scores between 2.60 and 2.93 the MSP. Cut-off values for reliable species determin- were observed, whereas B. pseudomallei were recognized ation cannot be theoretically calculated and have to be with scores in the range from 2.57 to 2.92. The top- determined empirically. According to the manufacturer, ranking hit of the hit-list correctly indicated the species experience has shown that scores exceeding 2.0 will of all queried samples. Scores of all top-ranking hits allow reliable genus identification and species identifica- exceeded 2.8. Construction of a score-based dendrogram tion in the majority of cases. Scores calculated for all of B. mallei and B. pseudomallei samples (Figure 2) with spectra of the custom reference set among them are MALDI Biotyper software resulted in the expected clus- summarized in Figure 1. In the hit lists of all tested spe- tering of the two species. Interestingly, the B. pseudo- cimen, the highest-ranking entry represented the same mallei type strain ATCC 23343 separated notably from species as the tested specimen, indicating that, within other B. pseudomallei representatives. This was at least the given database, the standard MALDI Biotyper identi- in part caused by the appearance of two series of masses fication procedure reliably allows the determination of between 5,000 and 5,084 Da and 8,500 and 8,565 Da Burkholderia species including the differentiation be- which were not detected in any of the other samples tween B. mallei and B. pseudomallei. Even though spe- (Figure 3). The observation of multiple mass differences cies identification was correct in all cases, the of 14 Da in these series suggests that they were caused distribution of scores in Figure 1 gave rise to concern by multiple methylations being specific for this strain. about the reliability of the discrimination of the three The mass series reproducibly appeared in all single spec- members of the Pseudomallei group: B. thailandensis tra used to calculate the MSP of the B. pseudomallei produced relatively high scores with some of the B. mal- strain ATCC 23343 and were also observed in independ- lei and B. pseudomallei samples, and B. pseudomallei ent replicates of the spectra with a freshly cultivated spe- generally produced relatively high scores with B. mallei. cimen. The identity of the modified molecule is Therefore, a set of B. mallei and B. pseudomallei sam- unknown. A dendrogram was constructed from the MSP ples was additionally cultivated and processed in two dif- of the B. mallei and B. pseudomallei strains listed in ferent laboratories and queried using the custom Table 1 and the Burkholderia, Chromobacterium, and reference set as database in order to challenge the iden- Rhodococcus species from Table 2 which were added tification procedure. It is known that cultivation condi- from the MALDI Biotyper database (Figure 4). As tions can influence the outcome of ICMS experiments. expected, score-based distances between B. mallei and In an interlaboratory comparison that was performed in B. pseudomallei were smaller than between the other three laboratories with B. thailandensis we had observed Burkholderia species and B. mallei/B. pseudomallei and that cultivation on different growth media (Columbia 5% B. thailandensis formed a distinct group which was Sheep Blood agar (CSB), chocolate agar, and McConkey separated from the other species of the Burkholderia agar) and different cultivation periods (24, 48 and 72 h) genus. had a notable influence on the scores in the identifica- The distance relations of B. mallei and B. pseudomallei tion procedure (data not shown). To avoid any variance were further analysed after transfer of the mass lists into caused by differing growth conditions, all B. mallei and statistical programming language R. Based on the mass B.pseudomallei were grown on CSB and the cultivation alignment, a cluster analysis was performed, a distance period of 48 h was strictly observed. matrix was calculated, and the distances within and be- tween the B. mallei and B. pseudomallei strains were Discrimination of B. mallei and B. pseudomallei calculated. To test the influence of the peak intensities Scores between B. mallei samples listed in Table 1 ran- on the clustering behavior, cluster analysis was per- ged between 2.56 and 2.94, whereas those between B. formed with the quantitative and qualitative data. For pseudomallei samples ranged between 2.25 and 2.89. For the latter purpose the quantitative alignment containing B. mallei samples, the score range over 2.72 was com- the intensities of every mass peak was transformed into pletely reserved for correct species assignments and the a qualitative binary table by marking the absence or top scores of all isolates reached this threshold. Due to presence of a mass with 0 and 1, respectively. From both the stronger variation of B. pseudomallei, such a well- tables, distance matrices were calculated and visualized defined threshold for correct species assignments could as Sammon-plots (Figure 5). For qualitative and quanti- not be defined for this species. tative data the average normalized distances between B. As MSP will usually not be constructed for routine mallei strains were smaller than between B. pseudomal- samples, the identification procedure with MALDI lei strains (0.57 vs. 0.73 for the binary data and 0.46 vs. Karger et al. BMC Microbiology 2012, 12:229 Page 5 of 15 http://www.biomedcentral.com/1471-2180/12/229

A Burkholderia mallei B. mallei 34 B. mallei B. mallei 32 B. mallei B. mallei NCTC 120 B. mallei B. mallei NCTC 10260 B. mallei B. mallei NCTC 10230 B. mallei NCTC 10247 B. mallei B. mallei 235 B. mallei 237 B. mallei M1 B. mallei Mukteswar B. mallei B. mallei 242 B. mallei B. mallei ATCC 23344T ATCC B. mallei B. mallei SVPM1 B. mallei B. mallei M2 B. mallei 7 Dubai B. mallei Zagreb B. mallei B. mallei Bogor B. mallei score 123

BCBurkholderia pseudomallei other Burkholderia species 0490 B. pseudomallei PITT 225A PITT B. pseudomallei EF 15660 B. pseudomallei H05410- B. pseudomallei B. pseudomallei PITT 521 PITT B. pseudomallei 5691 PITT B. pseudomallei 120107RR0019 B. pseudomallei B. multivorans DSM 13243 B. multivorans B. cenocepacia ATCC BAA-245 ATCC B. cenocepacia 700388 ATCC B. thailandensis B. stabilis DSM 16586 B. stabilis B. pseudomallei ATCC 23343T ATCC B. pseudomallei B. dolosa DSM 16088 B. dolosa B. pseudomallei NCTC 1688 B. pseudomallei B. pseudomallei 03-04448 B. pseudomallei 03-04450 B. pseudomallei B. ambifaria DSM 16087 B. ambifaria 29195 ATCC B. glathei 3 score score 12 123

Figure 1 Summary of the MALDI Biotyper queries with the reference spectrum set. The three panels summarize the score-oriented hit lists that the thirty-four strains of the custom reference set produced when queried against the reference spectrum set plus all representatives of the Burkholderia genus present in the MALDI Biotyper reference database. The three panels represent queries of B. mallei (A), B. pseudomallei (B) and other members of the B. genus (C). Filled circles, squares and open circles indicate scores produced by database entries representing B. mallei, B. pseudomallei or any of the other species in the reference database. Note that for all samples the highest ranking hit represents a member of the respective Burkholderia species.

0.71 when peak intensities of the spectra were included) the species and the distances between the species was confirming the score-based clustering in Figure 2 that calculated. Transition from qualitative to quantitative suggests a higher variation among B. pseudomallei than data showed slight improvement (0.82 vs. 0.74) in the among B. mallei strains. As a measure for the separation species separation indicating that peak intensities are of the two species, the weighted ratio between the dis- relevant for the discrimination of the two species and tances of B. mallei and B. pseudomallei strains within should not be neglected. Cluster analysis with the Karger et al. BMC Microbiology 2012, 12:229 Page 6 of 15 http://www.biomedcentral.com/1471-2180/12/229

B.pseudomallei PITT 5691

B.pseudomallei NCTC 1688

B.pseudomallei ATCC 23343

B.pseudomallei PITT 521

B.pseudomallei H05410-0490

B.pseudomallei EF 15660

B.pseudomallei PITT 225

B.pseudomallei 03 04450

B.pseudomallei 120107RR0019

B.pseudomallei 03 04448

B.mallei Mukteswar

B.mallei NCTC 120 Lister

B.mallei NCTC 10260

B.mallei 34

B.mallei 242

B.mallei NCTC 10247

B.mallei 237

B.mallei Dubai 7

B.mallei Zagreb

B.mallei Rotz 7(SVP)

B.mallei Bogor

B.mallei ATCC 23344

B.mallei M2

B.mallei M1

B.mallei NCTC 10230

B.mallei 32

B.mallei 235

1000 800 600 400 200 0

Distance Level Figure 2 Dendrogram obtained for Burkholderia mallei and Burkholderia pseudomallei strains. Spectrum-based distances between members of the B. mallei species are usually smaller than between representatives of B. pseudomallei. quantitative data using the k-means algorithm indicated spectra, most probably because of the multiple modifica- the presence of two clusters which were congruent with tions shown in Figure 3. the two Burkholderia species whereas cluster analysis As some B. mallei and B. pseudomallei specimen from based on the qualitative data failed to do so. On basis of the reference spectrum set produced quite high scores the qualitative data, which weights every mass equally with the respective other species, it was essential to test for the calculation of the distance, B. pseudomallei the practicability of the custom reference set in a routine ATCC 23343 was notably separated from all other laboratory setting with samples prepared in a different Karger et al. BMC Microbiology 2012, 12:229 Page 7 of 15 http://www.biomedcentral.com/1471-2180/12/229

5000 2500 4987.21 8629.68

4000 2000

3000 1500 8448.49 A

2000 1000

1000 500

0 0

2000 2500 4987.34 8447.61 8629.02 2000 1500

1500 1000 B

1000

500 500

0 0 4000

3000 8497.72 4987.34 8512.10 8525.92 5042.66 2500 3000 8466.96 8539.91 5057.08 8484.64 5028.57 5070.95 5085.28

2000 5013.99 8554.17 5098.94 8567.87 2000 8453.60 C

1500 5000.01 5113.26 8581.83

1000 5126.70 8632.31 1000 8595.76 8610.87 500

4980 5000 5020 5040 5060 5080 5100 5120 8450 8500 8550 8600 m/z m/z Figure 3 Unique modification patterns found for two proteins of B. pseudomallei ATCC23343T. Two regions of representative spectra of the three strains Burkholderia (B.) mallei Bogor (panel A), B. pseudomallei NCTC 1688 (panel B) and B. pseudomallei ATCC 23343 (panel C) are shown. Two striking series of multiple peaks with m/z distances of 14 Da were observed in B. pseudomallei ATCC 23343 but in no other of the tested isolates.

laboratory. The panel of samples used for this test (Table 3, Using the complete custom reference set as database a the ‘test set’) only partially overlapped with the custom number of misclassifications of the test set isolates oc- reference set (Table 1) so that not only inter-laboratory curred. Considering the distribution of scores (Figure 1) variation was tested but also the ability of the custom and the distance relations between B. mallei and B. reference set to discriminate newly appearing isolates like pseudomallei (Figure 5), this was not unexpected and those from a glanders outbreak in the United Arabic Emi- obviously a consequence of the indiscriminate inclusion rates in 2004. of all available B. mallei and B. pseudomallei samples Karger et al. BMC Microbiology 2012, 12:229 Page 8 of 15 http://www.biomedcentral.com/1471-2180/12/229

Table 2 Bacteria investigated for specificity testing Table 2 Bacteria investigated for specificity testing Species Strain (Continued) Burkholderia (B.) ambifaria LMG 11351 R. equi DSM 46064 B. ambifaria DSM 16087 T R. equi 559 LAL T B. anthina DSM 16086 T type strain. B. anthina LMG 16670 List of bacteria to be differentiated from Burkholderia mallei and Burkholderia pseudomallei using MALDI-TOF mass spectrometry. These bacteria include B. caledonica LMG 19076 T closely related bacteria, possible sample contaminants, bacteria with very similar clinical presentation and other relevant bacteria. MSP reference spectra B. caribensis* DSM 13236 T were constructed for the species indicated with an asterisk (*); all other B. cenocepacia LMG 12614 samples indicate isolates of the MALDI Biotyper database. B. cenocepathia* ATCC BAA-245 B. cepacia MB_7544_05 into the custom reference set. Classification could be B. cepacia DSM 11737 substantially improved by selecting combinations of iso- lates of B. mallei and B. pseudomallei to form a dedi- B. cepacia 18875_1 CHB cated reference set which is optimized for the B. cepacia DSM 9241 discrimination of the two species. To screen the B. cepacia DSM 50181 complete custom reference set of B. mallei and B. pseu- B. cepacia LMG 2161 domallei for appropriate combinations of isolates, the T B. cepacia* DSM 7288 outcome of a database query was simulated with all per- B. cepacia ATCC 25416 T mutations of up to four members of each species. The B. dolosa DSM 16088 smallest reference group yielding error-free results was B. fungorum LMG 20227 T composed of two B. mallei (M1, NCTC10247) and three B. gladioli Wv22575 CHB B. pseudomallei (EF15660, PITT 225A, NCTC01688) B. gladioli DSM 4285 T isolates which are highlighted by an asterisk in Table 1. B. glathei DSM 50014 T Not surprisingly, these isolates located close to the cen- ters of their respective species in the Sammon plot B. glumae DSM 9512 T visualization of the distance matrix (Figure 5). B. multivorans LMG 14293 Finally, multivariate statistics on basis of the four dif- T B. multivorans DSM 13243 ferent statistical approaches (Genetic Algorithm, Sup- T B. phenazinium DSM 10684 port Vector Machine, Supervised Neural Network, B. phymatum LMG 21445 T Quick Classifier) available in ClinProTools 3.0 showed B. plantarii DSM 9509 T that B. mallei and B. pseudomallei could be well sepa- B. pyrrocinia DSM 10685 T rated with cross validation results ranging between B. pyrrocinia LMG 14191 T 98.95% and 100.00% (data not shown). Principal Compo- B. sacchari LMG 19450 T nent Analysis (PCA) carried out with ClinProTools 3.0 B. stabilis LMG 14294 T (Figure 6) further confirmed the separation of both spe- cies and also the broader distribution of B. pseudomallei B. stabilis DSM 16586 T in comparison with B. mallei. B. terricola LMG 20594 T B. thailandensis DSM 13276 T Identification of taxon-specific biomarker ions B. thailandensis* ATCC 700388 Mass spectra of the reference spectrum set were analysed B. tropica DSM 15359 T T for species-specific masses which may be used for species B. tuberum LMG 21444 identification independent of the score values considered T B. vietnamiensis LMG 10929 so far. For that purpose the mass lists of the MSP gener- B. xenovorans LMG 21463 T ated with MALDI Biotyper software were evaluated in de- Chromobacterium (C.) subtsugae DSM 17043 T tail. An alignment of all masses occurring in the spectra C. violaceum C49 MVO was constructed as a table in which every column repre- C. violaceum DSM 30191T sented the mass spectrum of a sample and every row the Rhodococcus (R.) equi DSM 1990 intensity of a mass occurring in a certain mass range. The R. equi DSM 20295 alignment contained a total of 350 masses. One of them, R. equi DSM 20307 T 4,410 Da, was found in the spectra of all 34 samples and also in every one of the single spectra underlying the R. equi DSM 43950 MSPs so that this mass can be considered a reliable indi- R. equi* DSM 44426 cator of the set of nine Burkholderia species that was Karger et al. BMC Microbiology 2012, 12:229 Page 9 of 15 http://www.biomedcentral.com/1471-2180/12/229

R.equi DSM 46064 R.equi DSM 46101 R.equi DSM 43950 R.equi DSM 20295 R.equi DSM 1990 R.equi DSM 20307T R.equi Cory_35 R.equi 559 Chr.violaceum C49 Chr.violaceum DSM 30191T Chr.subtsugae DSM 17043T B.tuberum LMG 21444T B.tropica DSM 15359T B.sacchari LMG 19450T B.glathei DSM 50014T B.phenazinium DSM 10684T B.xenovorans LMG 21463T B.fungorum LMG 20227T B.terricola LMG 20594T B.phymatum LMG 21445T B.caledonica LMG 19076T B.gladioli Wv22575 B.gladioli DSM 4285T B.plantarii DSM 9509T B.glumae DSM 9512T B.multivorans LMG 14293 B.multivorans DSM 13243T B.dolosa DSM 16088T B.cepacia MB_7544_05 B.pyrrocinia DSM 10685T B.vietnamiensis LMG 10929T B.cepacia DSM 11737 B.anthina LMG 16670 B.anthina DSM 16086T B.ambifaria LMG 11351 B.stabilis LMG 14294T B.pyrrocinia LMG 14191T B.cepacia LMG 2161 B.cenocepacia LMG 12614 B.stabilis DSM 16586T B.cepacia DSM 9241 B.cepacia DSM 7288T B.cepacia DSM 50181 B.ambifaria DSM 16087T B.thailandensis DSM 13276T B.pseudomallei PITT 5691 B.pseudomallei NCTC 1688 B.pseudomallei H05410-0490 B.pseudomallei EF15660 B.pseudomallei PITT 521 B.pseudomallei PITT 225 B.pseudomallei 03_04450 B.pseudomallei 120107RR0019 B.pseudomallei 03_04448 B.pseudomallei ATCC 23343 B.mallei NCTC 10230 B.mallei 32 B.mallei 235 B.mallei M2 B.mallei M1 B.mallei Mukteswar B.mallei NCTC 10260 B.mallei NCTC 10247 B.mallei 34 B.mallei NCTC 120 Lister B.mallei 242 B.mallei 237 B.mallei Dubai 7 B.mallei ATCC 23344 B.mallei M1 B.mallei Zagreb B.mallei Bogor 1000 800 600 400 200 0 Distance Level Figure 4 Spectrum-based dendrogram representing Burkholderia mallei, Burkholderia pseudomallei, and other relevant bacteria. The dendrogram was constructed based on the MALDI Biotyper scores. Note that distances between B. mallei and B. pseudomallei isolates are small compared to the distances of other B. species. B. mallei/B. pseudomallei and B. thailandensis separate as distinct group from the other species of the B. genus. Karger et al. BMC Microbiology 2012, 12:229 Page 10 of 15 http://www.biomedcentral.com/1471-2180/12/229

A B

B. pseudomallei PITT 5691 B. pseudomallei ATCC 23343T

B. pseudomallei ATCC 23343T 6

B. mallei Mukteswar B. pseudomallei NCTC 1688

B. mallei ATCC 23344T B. mallei NCTC 10260 B. pseudomallei EF 15660 B. pseudomallei PITT 5691 B. mallei Zagreb B. mallei 237 B. pseudomallei H05410-0490 0.5 1.0 B. mallei M1 B. mallei 32 B. mallei Bogor B. mallei Dubai 7

B. mallei 237 B. pseudomallei 120107RR0019 B. mallei 34 B. mallei Mukteswar B. mallei 235 B. mallei NCTC 10247 B. pseudomallei 03-04450 B. pseudomallei PITT 521 0.0

B. mallei NCTC 120 B. mallei NCTC 120 B. pseudomallei 03-04448 B. mallei M2 B. pseudomallei PITT 225A B. mallei M2 B. mallei 242 B. mallei M1 B. mallei NCTC 10260 B. mallei NCTC 10247 B. mallei SVPM1 B. mallei SVPM1 B. mallei 32 B. pseudomallei NCTC 1688 -0.5 B. mallei Bogor B. mallei 34 B. mallei 235 B. pseudomallei 03-04448

-2B. mallei 0242 2 4 B. pseudomallei 03-04450 B. mallei Zagreb B. pseudomallei 120107RR0019 -1.0 B. pseudomallei H05410-0490 B. mallei Dubai 7 B. pseudomallei EF 15660 B. mallei NCTC 10230 -4 B. mallei NCTC 10230

B. pseudomallei PITT 225A

B. mallei ATCC 23344T -1.5 B. pseudomallei PITT 521

-4 -2 0 2 4 6 -0.5 0.0 0.5 1.0 1.5 Figure 5 Sammon representation of the spectrum-based distance relations of B. mallei and B. pseudomallei. Diagrams A and B were calculated from qualitative or quantitative distance matrices derived from the mass alignment of the spectra, respectively. Members of the dedicated reference spectrum set for the discrimination of B. mallei and B. pseudomallei are underlined. Sammon representations allow visualising distance matrices in a two-dimensional plot with minimized distortion. analysed (B. mallei, B pseudomallei, B, thailandensis, B. Discussion ambifaria, B. cenocepacia, B. dolosa, B. glathe, B. multi- In recent years MALDI-TOF MS has been introduced in vorans, B. stabilis). Seven more masses (3,655 [doubly microbiological laboratories as a time saving diagnostic charged 7,309], 5,195, 6,551, 7,169, 7,309, 8,628 and approach supplementing morphological, biochemical, 9,713 Da) were present in all B. mallei and B. pseudomal- and molecular techniques for identification of microbes lei samples but also in one or more of the other Burkhol- [23]. In several studies the comparability with conven- deria species. Considering the close relation of B. tional identification procedures was assessed with gener- thailandensis with B. mallei and B. pseudomallei, mass ally good correlation, but discordances were seen on the 9,713 Da is of interest, which was specific for all B. mallei, species and even on the genus level [24,25]. This prote- B. pseudomallei, and B. thailandensis samples, i.e. the omic profiling approach was successfully applied in rou- Pseudomallei group. Finally, 6,551 Da was present in all B. tine identification of bacterial isolates from blood mallei and B. pseudomallei samples but in none of the culture with the exception of polymicrobial samples and other species, making it an effective discriminator between streptococci [26]. The identification of Burkholderia spp. the B. mallei/pseudomallei group and the other represen- and other non-fermenting bacteria using MALDI-TOF tatives of the genus Burkholderia. Concerning the distinc- MS was investigated in cystic fibrosis (CF) patients as tion of B. mallei and B. pseudomallei, statistical analysis Burkholderia spp. (mainly of the cepacia-complex) cause with ClinProTools 3.0 software revealed a number of a relevant number of life-threatening infections in these masses with significant class separation between the two patients [27-29]. It was demonstrated that MALDI-TOF species based on peak intensity. Most significant separ- MS is a useful tool for rapid identification in the routine ation could be obtained based on the masses 7,553 and laboratory. B. pseudomallei can be the cause of melioid- 5,794 which differ significantly in intensity between the osis in CF patients and travelers to tropical regions, but two species. this bacterium and the closely related species B. mallei Karger et al. BMC Microbiology 2012, 12:229 Page 11 of 15 http://www.biomedcentral.com/1471-2180/12/229

Table 3 Bacteria used to test the reliability of ICMS-based discrimination of Burkholderia mallei and Burkholderia pseudomallei Species Strain designation Score B. mallei 32 2.470 34 2.475 237 2.189 B.mallei 242 2.550 ATCC 23344T 2.382 Bogor 2.522 B.pseudomallei Mukteswar 2.554 Zagreb 2.472 NCTC 120 2.478 NCTC 10260 2.092 NCTC 10247 2.325 NCTC 10230 1.960 05-767 2.329 Figure 6 Principal component analysis of spectra derived from B. mallei and B. pseudomallei. Principle Component Analysis of ten 05-762 2.515 strains of B. mallei and ten strains of B. pseudomallei, respectively. 05-2316 2.496 The unsupervised statistical analysis separates both species based on Dubai3-10, 14-17* 2.437 - 2.630 the three major principle components. While B. mallei form a relatively uniform cluster, significant diversity can be observed for B. B. pseudomallei EF 15660 2.692 pseudomallei. Analysis of the spectra from the specimens in Table 1 NCTC 1688 2.489 yielded very similar results (data not shown). 06-2372 2.588 06-2377 2.621 species was not successful. Obviously, more complicated 06-2379 2.427 mass signatures are required for this purpose and, as we 06-2388 2.603 could show after separate statistical evaluation of quali- 06-2393 2.328 tative and quantitative data, peak intensities also play a 06-2395 2.633 crucial role for the discrimination of B. mallei and B. 06-772 2.379 pseudomallei. However, group-specific masses like 9,713 Da, standing for the Pseudomallei complex (B. *B. mallei isolates from several horses isolated during the glanders outbreak in UAE 2004. mallei/B. pseudomallei/B. thailandensis) or 6,551, exclu- List of strains used to evaluate the reliability of ICMS-based discrimination of B. sively found in B. mallei and B. pseudomallei may be of mallei and B. pseudomallei using a dedicated set of reference strains. Column ‘Score’ designates the score value of the top-ranking hit in the dedicated use for the discrimination of these three species. database, which in all cases represented the same species as the tested For the identification of B. mallei and B. pseudomallei T sample. ( , typestrain). samples under routine laboratory conditions, it was ne- cessary to reduce the reference spectrum set to avoid was not included in previous MALDI-TOF MS studies misclassifications. Interestingly, the reference spectrum [18-22,30,31]. Natural catastrophes like the tsunami in set optimized for spectrum-based discrimination neither Indonesia (2004) and occasional flooding in other trop- contained the type strain ATCC 23344T (B. mallei) nor ical regions resulted in elevated incidence of melioidosis ATCC 23343T (B. pseudomallei). One reason for the ex- and several cases among travelers and tourists [32-36]. clusion of ATCC 23343 could be the occurrence of two B. mallei and B. pseudomallei are biological agents peak series with repeating mass increments of 14 Da which further underlines the need for rapid detection most probably representing polymethylated proteins. tools. Identification of Burkholderia ssp. and distinction This strain has been shown to have unique immuno- of B. mallei and B. pseudomallei from other species was logical features. In an immunization experiment with a feasible. This was also true for the closely related bacter- panel of 14 B. pseudomallei strains, ATCC 23343 ium B. thailandensis. All strains grew well within induced monoclonal antibodies in mice which did not 48 hours and could then be readily prepared for cross-react with any of the other B. pseudomallei MALDI-TOF MS. strains [37]. These peculiarities may indicate that this Due to the close relationship of B. mallei and B. pseu- type strain has been genetically modified by frequent domallei, it was not surprising that the search for subcultivation or misuse of media. To our knowledge, species-identifying biomarker ions discriminating these similar modifications which may have an impact on Karger et al. BMC Microbiology 2012, 12:229 Page 12 of 15 http://www.biomedcentral.com/1471-2180/12/229

classification of bacteria have not been reported to-date. set of strains enlisted in Table 3, referred to as ‘test set,’ These series were specific for the isolate and also for were recorded with an Autoflex mass spectrometer (Bru- two molecules within the observed mass range. ker) in a second laboratory and queried against the refer- ence set to test for robustness and inter-laboratory Conclusions variation. In this study we have demonstrated that isolates of the closely related species B. mallei and B. pseudomallei can Intact cell mass spectrometry (ICMS) be identified using MALDI-TOF MS. Dangerous and Samples were prepared as described previously [16,40]. cumbersome handling under BSL 3 conditions can be Briefly, the bacteria were cultivated under BSL 3 condi- minimized by inactivation of the isolates with ethanol tions on a nutrient blood agar containing 3% glycerol at and subsequent MALDI-TOF MS analysis that requires 37°C for 48 hours. Specimens from single colonies were much less time than nucleic acid amplification methods thoroughly suspended in 300 μl water and precipitated [38]. The reference spectra exhibited a higher homogen- by addition of 900 μL ethanol (98% v/v). This treatment eity among B. mallei than among B. pseudomallei. The inactivated the bacteria as was demonstrated by growth type strain of B. pseudomallei ATCC 23343 was isolated controls and the specimens could be further tested decades ago and separated from the other B. pseudomal- under BSL 1 conditions. After sedimentation for five lei specimens in the dendrograms which is probably due minutes at 10,000 g min-1 the supernatant was carefully to polymethylation as indicated by two intensive series removed and the sediment suspended in 50 μL of 70% of mass increments of 14 Da. To our knowledge, this is (v/v) formic acid. After mixing with 50 μL acetonitrile, the first report of such a modification in whole cell the suspension was centrifuged as described above and MALDI-TOF MS spectra of microorganisms. As the supernatant transferred to a fresh tube. 1.5 μL of the expected for closely related species, especially when one extract was spotted onto a steel MALDI target plate and of them, B. pseudomallei, displays the broad distribution allowed to dry at ambient temperature. Finally, the dried of MALDI-phenotypes that was observed, the identifica- extract was overlaid with 2 μL of a saturated solution of tion process requires the selection of spectra from repre- α-Cyano-4-hydroxycinnamic acid in 50% acetonitrile/ sentatives of the centers of their respective distance 2.5% trifluoroacetic acid as matrix and was again allowed distributions. Uncritical inclusion of all available samples to dry. A custom-made database of reference spectra as references in a library was counterproductive for the was generated using the MALDI Biotyper software (ver- identification process. Only by selecting an appropriate sion 2.0, Bruker Daltonik GmbH, Bremen, Germany) fol- set of reference spectra (Table 3) it was possible to iden- lowing the guidelines of the manufacturer. Each sample tify all strains. This underlines the need for careful cur- was spotted onto six target spots of a steel MALDI tar- ation of reference spectra databases used for the get plate. Spectra were acquired with an UltraflexTM I identification of microorganisms. instrument (Bruker Daltonik GmbH) in the linear posi- tive mode in the range of 2,000 to 20,000 Da. Acceler- Methods ation Voltage was 25 kV and the instrument was Bacterial strains calibrated in the range between 3,637.8 and 16,952.3 Da A comprehensive collection of B. mallei and B. pseudo- with Bacterial Test Standard calibrant (BTS, Bruker Dal- mallei strains, referred to as the ‘reference set,weretested’ tonik GmbH). Four single mass spectra with 500 shots (Table 1) and compared with spectra of closely related and each were acquired from each spot and a reference other clinically relevant bacteria (Table 2) included in the spectrum calculated from the 24 single spectra. Refer- MALDI Biotyper Reference Library (version 3.0, Bruker ence spectra contained the parameters peak mass and Daltonics, Bremen, Germany). Strain identity was con- intensity and additional information on the reproducibil- firmed using Gram staining, motility testing, and real-time ity of the mass peaks, i.e. the frequency of occurrence of PCR assays targeting fliC and fliP as described previously every peak in the underlying 24 single spectra. Reference [11,12], and a species-specific DNA-microarray [39]. spectra were generated within the mass range of 2,000 Strains were obtained from the Friedrich-Loeffler-Institut, to 20,000 Da with the default parameter settings in the Jena, Germany and the Bundeswehr Institute of Micro- MALDI Biotyper software. The number of peaks was biology in Munich, Germany. Strain Dubai 7 was kindly limited to 100 per reference spectrum and all peaks of a provided by the Central Veterinary Reference Laboratory, reference spectrum were normalized to the most intense Dubai, UAE. Some strains originated from the Robert peak with an intensity of 1.0. The minimum frequency Koch Institute in Berlin, Germany, that coordinated a pro- of occurrence within the 24 single spectra was set to ject of the European Union for the “Establishment of 50% for every mass. Peaklists of reference spectra were Quality Assurances for Detection of Highly Pathogenic exported for further evaluation in the statistical pro- Bacteria of Potential Bioterrorism Risk”. Spectra from the gramming language R. Karger et al. BMC Microbiology 2012, 12:229 Page 13 of 15 http://www.biomedcentral.com/1471-2180/12/229

To test the inter-laboratory variation and the robust- Statistical analysis with ClinProTools software ness of the classification by using MALDI Biotyper soft- Raw spectra from the specimens in Table 3 were ware, a set of B. mallei and B. pseudomallei test samples imported into ClinProTools 3.0 software for statistical from a second laboratory (Table 3) was queried against analysis. Each species was represented by 20 to 24 the reference spectra set described above. These spectra spectra to cover measurement variability. The multiple were recorded at the Bundeswehr Institute of Microbiol- spectra of multiple species were imported as a “class” ogy with an Autoflex mass spectrometer (Bruker Dalto- for the respective species. ClinProTools preformed a nik GmbH, Bremens). Spectra were generated for the normalization and recalibration of mass spectra before test set in the same way as for the reference set. A query further analysis, thereby reducing measurement variabil- of all test samples was performed and the resulting ity effects significantly. Peak picking was performed scores were transferred into a data matrix, the ‘score based on the overall average spectrum over the whole matrix’, in which every column represented a member of mass range (signal to noise threshold of 5). Further spec- the reference set and every line a test sample. The power tra processing parameters were: baseline correction of certain combinations of representatives of the two (convex hull), resolution (300 ppm), smoothing (Savitzky classes within the reference set to discriminate samples Golay, 5 cycles with 2 m/z width), of the test set was tested as follows: the columns repre- Multivariate statistical analyses were performed using senting the combination of reference spectra to be eval- the four supervised algorithms and PCA which are uated were selected from the score matrix and every implemented in ClinProTools. For the Genetic Algo- member of the test set was classified by assigning it to rithm, models with maximum 5 peaks and 50 genera- the class of the member of the reduced reference set tions were calculated and k-nearest neighbor (kNN) with the highest score. The number of correct and in- classification was performed with 5 neighbors. Also for correct assignments was then calculated for the test set. Support Vector Machine the maximum number of peaks This procedure simulates a MALDI Biotyper query with was set to 5 and kNN classification was performed with a reduced number of spectra in the reference database. 5 neighbors. Supervised Neural Network was calculated with automated optimization of peak number, maximum Transfer of data into statistical programming language R 25. For the Quick Classifier, a maximum number of dif- Peak lists of the reference spectra generated by the ferentiating peaks of 25 was allowed; selection of peaks MALDI Biotyper software were converted into a format was based on ranking in t-test. For PCA, “level” scaling compatible with the R-package caMassClass which offers was selected. a number of functions to evaluate mass spectra [41] for further statistical analysis in R version 2.10.1. available at Authors’ contributions the R-project homepage [42]. Peak lists were aligned by AK performed MALDI-TOF MS experiments, data analysis and participated in drafting the manuscript. RS worked in the BSL3 laboratory, performed the msc.peaks.align command of caMassClass and trans- MALDI-TOF MS experiments and data analysis. MZ developed R-scripts and formed into a binary mass table where rows represented participated in the mathematical analysis of mass spectra and in solving all unique masses of the aligned spectra set and every classification problems. MCE coordinated the work in the BSL3 laboratory, performed cultivation and PCR assays. BB performed MALDI-TOF MS column represented the spectrum of one sample. The experiments and data analysis. FM worked in the BSL3 laboratory, performed size of the mass ranges defining a unique peak in the cultivation and PCR assays. TM performed MALDI-TOF MS experiments and alignment, designated as bin size, was restricted to a data analysis. MK performed data analysis and statistical examination. HCS worked in the BSL3 laboratory, performed cultivation and PCR assays, and maximum of 2,000 ppm. Among other features, the al- critically reviewed the manuscript. HN critically reviewed the manuscript. HT gorithm of the msc.peaks.align command minimizes the participated in the design of the study, coordinated the experiments, and bin size in the given range, maximizes the space between participated in drafting the manuscript. MK and TM are employees of Bruker Daltonik GmbH, the manufacturer of the MALDI Biotyper system used in this bins and ensures that no two peaks of the same study. All authors read and approved the final manuscript. spectrum are in the same bin. For the calculation of qualitative data, the presence of the respective mass in Acknowledgements the spectrum of a sample was marked with 1, absence We are grateful to Gabi Echle, Katja Fischer, Michaela Ganss, and Robert with 0, i.e. all mass intensities were removed. These Schneider for their excellent technical assistance. This work was supported by the EU, EAHC Agreement - No 2007 204. tables were the basis for the calculation of distances (R-routine ‘dist’, parameter ‘binary’ for the distance Author details 1 measure) which were used for the construction of clado- Friedrich-Loeffler-Institut, Federal Research Institute for Animal Health, Institute of Molecular Biology, Südufer 10, Greifswald-Insel Riems D-17493, grams, Sammon plots [43], and k-means cluster analysis Germany. 2Bundeswehr Institute of Microbiology, Neuherbergstrasse 11, using the R-routines ‘hclust’ (parameter ‘ward’ for the Munich D-80937, Germany. 3Working Group Biomathematics, agglomeration method) [44], ‘sammon’ (used with default Friedrich-Loeffler-Institut, Federal Research Institute for Animal Health, ‘ ’ Südufer 10, Greifswald-Insel Riems D-17493, Germany. settings) and kmeans (three initial cluster centers, max- 4Friedrich-Loeffler-Institut, Federal Research Institute for Animal Health, imum of 100 iterations, Hartigan-Wong algorithm [45]). Institute of Bacterial Infections and Zoonoses, Naumburger Str. 96 a, Jena Karger et al. BMC Microbiology 2012, 12:229 Page 14 of 15 http://www.biomedcentral.com/1471-2180/12/229

07743, Germany. 5Bruker Daltonik GmbH, Fahrenheitstr. 4, Bremen 28359, 19. Corral DM, Coates AL, Yau YC, Tellier R, Glass M, Jones SM, Waters VJ: Germany. Burkholderia pseudomallei infection in a cystic fibrosis patient from the Caribbean: a case report. Can Respir J 2008, 15(5):237–239. Received: 22 March 2012 Accepted: 25 September 2012 20. Holland DJ, Wesley A, Drinkovic D, Currie BJ: Cystic Fibrosis and Published: 10 October 2012 Burkholderia pseudomallei Infection: An Emerging Problem? Clin Infect Dis 2002, 35(12):e138–e140. 21. Schulin T, Steinmetz I: Chronic melioidosis in a patient with cystic fibrosis. References J Clin Microbiol 2001, 39(4):1676–1677. 1. Dance DA: Melioidosis: the tip of the iceberg? Clin Microbiol Rev 1991, 22. Visca P, Cazzola G, Petrucca A, Braggion C: Travel-associated Burkholderia 4(1):52–60. pseudomallei infection (Melioidosis) in a patient with cystic fibrosis: a 2. Deris ZZ, Hasan H, Siti Suraiya MN: Clinical characteristics and outcomes of case report. Clin Infect Dis 2001, 32(1):E15–E16. bacteraemic melioidosis in a teaching hospital in a northeastern state of 23. Seng P, Rolain JM, Raoult D, Brouqui P: Detection of new Malaysia: a five-year review. J Infect Dev Ctries 2010, 4(7):430–435. Anaplasmataceae in the digestive tract of fish from southeast Asia. Clin 3. Sprague LD, Neubauer H: Melioidosis in animals: a review on Microbiol Infect 2009, 15(Suppl 2):88–90. epizootiology, diagnosis and clinical presentation. J Vet Med B Infect Dis 24. Ferreira L, Vega S, Sanchez-Juanes F, Gonzalez M, Herrero A, Muniz MC, Vet Public Health 2004, 51(7):305–320. Gonzalez-Buitrago JM, Munoz JL: [Identifying bacteria using a matrix- 4. Estes DM, Dow SW, Schweizer HP, Torres AG: Present and future assisted laser desorption ionization time-of-flight (MALDI-TOF) mass therapeutic strategies for melioidosis and glanders. Expert Rev Anti Infect spectrometer. Comparison with routine methods used in clinical Ther 2010, 8(3):325–338. microbiology laboratories]. Enferm Infecc Microbiol Clin 2010, 28(8):492–497. 5. Dance DA, Wuthiekanun V, Naigowit P, White NJ: Identification of 25. Risch M, Radjenovic D, Han JN, Wydler M, Nydegger U, Risch L: Comparison Pseudomonas pseudomallei in clinical practice: use of simple screening of MALDI TOF with conventional identification of clinically relevant tests and API 20NE. J Clin Pathol 1989, 42(6):645–648. bacteria. Swiss Med Wkly 2010, 140:w13095. 6. Inglis TJ, Chiang D, Lee GS, Chor-Kiang L: Potential misidentification of 26. La Scola B, Raoult D: Direct identification of bacteria in positive blood Burkholderia pseudomallei by API 20NE. Pathology 1998, 30(1):62–64. culture bottles by matrix-assisted laser desorption ionisation time-of-flight 7. Lowe P, Engler C, Norton R: Comparison of automated and mass spectrometry. PLoS One 2009, 4(11):e8041. nonautomated systems for identification of Burkholderia pseudomallei. 27. Degand N, Carbonnelle E, Dauphin B, Beretti JL, Le Bourgeois M, Sermet- J Clin Microbiol 2002, 40(12):4625–4627. Gaudelus I, Segonds C, Berche P, Nassif X, Ferroni A: Matrix-assisted laser 8. Samosornsuk N, Lulitanond A, Saenla N, Anuntagool N, Wongratanacheewin desorption ionization-time of flight mass spectrometry for identification S, Sirisinha S: Short report: evaluation of a monoclonal antibody-based of nonfermenting gram-negative bacilli isolated from cystic fibrosis latex agglutination test for rapid diagnosis of septicemic melioidosis. patients. J Clin Microbiol 2008, 46(10):3361–3367. AmJTrop Med Hyg 1999, 61(5):735–737. 28. Minan A, Bosch A, Lasch P, Stammler M, Serra DO, Degrossi J, Gatti B, Vay C, 9. Steinmetz I, Reganzerowski A, Brenneke B, Haussler S, Simpson A, White NJ: D'Aquino M, Yantorno O, et al: Rapid identification of Burkholderia Rapid identification of Burkholderia pseudomallei by latex agglutination cepacia complex species including strains of the novel Taxon K, based on an exopolysaccharide-specific monoclonal antibody. J Clin recovered from cystic fibrosis patients by intact cell MALDI-ToF mass Microbiol 1999, 37(1):225–228. spectrometry. Analyst 2009, 134(6):1138–1148. 10. Thibault FM, Valade E, Vidal DR: Identification and discrimination of 29. Vanlaere E, Sergeant K, Dawyndt P, Kallow W, Erhard M, Sutton H, Dare D, Burkholderia pseudomallei, B. mallei, and B. thailandensis by real-time PCR Devreese B, Samyn B, Vandamme P: Matrix-assisted laser desorption targeting type III secretion system genes. J Clin Microbiol 2004, ionisation-time-of of-flight mass spectrometry of intact cells allows rapid 42(12):5871–5874. identification of Burkholderia cepacia complex. J Microbiol Methods 2008, 11. Tomaso H, Pitt TL, Landt O, Al Dahouk S, Scholz HC, Reisinger EC, 75(2):279–286. Sprague LD, Rathmann I, Neubauer H: Rapid presumptive 30. Currie BJ: Melioidosis: an important cause of pneumonia in residents of identification of Burkholderia pseudomallei with real-time PCR and travellers returned from endemic regions. Eur Respir J 2003, assays using fluorescent hybridization probes. Mol Cell Probes 2005, 22(3):542–550. 19(1):9–20. 31. O'Carroll MR, Kidd TJ, Coulter C, Smith HV, Rose BR, Harbour C, Bell SC: 12. Tomaso H, Scholz HC, Al Dahouk S, Eickhoff M, Treu TM, Wernery R, Burkholderia pseudomallei: another emerging pathogen in cystic fibrosis. Wernery U, Neubauer H: Development of a 5'-nuclease real-time PCR Thorax 2003, 58(12):1087–1091. assay targeting fliP for the rapid identification of Burkholderia mallei in 32. Christenson B, Fuxench Z, Morales JA, Suarez-Villamil RA, Souchet LM: clinical samples. Clin Chem 2006, 52(2):307–310. Severe community-acquired pneumonia and sepsis caused by 13. Tomaso H, Scholz HC, Al Dahouk S, Pitt TL, Treu TM, Neubauer H: Burkholderia pseudomallei associated with flooding in Puerto Rico. Bol Development of 5' nuclease real-time PCR assays for the rapid Asoc Med P R 2003, 95(6):17–20. identification of the burkholderia mallei//burkholderia pseudomallei 33. Ciervo A, Mattei R, Cassone A: Melioidosis in an Italian tourist injured by complex. Diagn Mol Pathol 2004, 13(4):247–253. the tsunami in Thailand. J Chemother 2006, 18(4):443–444. 14. Giebel R, Worden C, Rust SM, Kleinheinz GT, Robbins M, Sandrin TR: 34. Nieminen T, Vaara M: Burkholderia pseudomallei infections in Finnish Microbial fingerprinting using matrix-assisted laser desorption ionization tourists injured by the December 2004 tsunami in Thailand. Euro Surveill time-of-flight mass spectrometry (MALDI-TOF MS) applications and 2005, 10(3):E050303 050304. challenges. Adv Appl Microbiol 2010, 71:149–184. 35. Svensson E, Welinder-Olsson C, Claesson BA, Studahl M: Cutaneous 15. Marklein G, Josten M, Klanke U, Muller E, Horre R, Maier T, Wenzel T, Kostrzewa melioidosis in a Swedish tourist after the tsunami in 2004. Scand J Infect M, Bierbaum G, Hoerauf A, et al: Matrix-assisted laser desorption ionization- Dis 2006, 38(1):71–74. time of flight mass spectrometry for fast and reliable identification of 36. Wuthiekanun V, Chierakul W, Rattanalertnavee J, Langa S, Sirodom D, clinical yeast isolates. JClinMicrobiol2009, 47(9):2912–2917. Wattanawaitunechai C, Winothai W, White NJ, Day N, Peacock SJ: 16. Sauer S, Freiwald A, Maier T, Kube M, Reinhardt R, Kostrzewa M, Geider K: Serological evidence for increased human exposure to Burkholderia Classification and identification of bacteria by mass spectrometry and pseudomallei following the tsunami in southern Thailand. J Clin Microbiol computational analysis. PLoSONE 2008, 3(7):e2843. 2006, 44(1):239–240. 17. Fernandez-Olmos A, Garcia-Castillo M, Morosini MI, Lamas A, Maiz L, Canton 37. Feng SH, Tsai S, Rodriguez J, Newsome T, Emanuel P, Lo SC: Development R: MALDI-TOF MS improves routine identification of non-fermenting of mouse hybridomas for production of monoclonal antibodies specific Gram negative isolates from cystic fibrosis patients. J Cyst Fibros 2012, to Burkholderia mallei and Burkholderia pseudomallei. Hybridoma (Larchmt) 11(1):59–62. 2006, 25(4):193–201. 18. Barth AL, de Abreu ESFA, Hoffmann A, Vieira MI, Zavascki AP, Ferreira AG, da 38. Lasch P, Nattermann H, Erhard M, Stammler M, Grunow R, Bannert N, Appel Cunha LG Jr, Albano RM, de Andrade Marques E: Cystic fibrosis patient B, Naumann D: MALDI-TOF mass spectrometry compatible inactivation with Burkholderia pseudomallei infection acquired in Brazil. J Clin method for highly pathogenic microbial cells and spores. Anal Chem Microbiol 2007, 45(12):4077–4080. 2008, 80(6):2026–2034. Karger et al. BMC Microbiology 2012, 12:229 Page 15 of 15 http://www.biomedcentral.com/1471-2180/12/229

39. Schmoock G, Ehricht R, Melzer F, Rassbach A, Scholz HC, Neubauer H, Sachse K, Mota RA, Saqib M, Elschner M: DNA microarray-based detection and identification of Burkholderia mallei, Burkholderia pseudomallei and Burkholderia spp. Mol Cell Probes 2009, 23(3–4):178–187. 40. Karger A, Ziller M, Bettin B, Mintel B, Schares S, Geue L: Determination of serotypes of Shiga toxin-producing Escherichia coli isolates by intact cell matrix-assisted laser desorption ionization-time of flight mass spectrometry. Appl Environ Microbiol 2011, 77(3):896–905. 41. Tuszynski J: caMassClass: Processing & Classification of Protein Mass Spectra (SELDI) Data; 2010. http://CRAN.R-project.org/package=caMassClass. 42. R Development Core Team: R: A language and environment for statistical computing. R Foundation for Statistical Computing. Vienna, Austria; 2009. 43. Sammon J: A non-linear mapping for data structure analysis. IEEE Trans Comp C 1969, 18:401–409. 44. Everitt B: Cluster analysis. London: Heinemann Educational Books; 1974. 45. Hartigan J, Wong M: A K-means clustering algorithm. Appl Statistics 1979, 28:100–108.

doi:10.1186/1471-2180-12-229 Cite this article as: Karger et al.: Rapid identification of Burkholderia mallei and Burkholderia pseudomallei by intact cell Matrix-assisted Laser Desorption/Ionisation mass spectrometric typing. BMC Microbiology 2012 12:229.

Submit your next manuscript to BioMed Central and take full advantage of:

• Convenient online submission • Thorough peer review • No space constraints or color figure charges • Immediate publication on acceptance • Inclusion in PubMed, CAS, Scopus and Google Scholar • Research which is freely available for redistribution

Submit your manuscript at www.biomedcentral.com/submit